Annals of Applied Sciences

A Novel Hybrid Model Based on Rule Learning and Dilated Convolution for Vehicle Collision Prediction

2024-01-08T21:05:05+00:00

Vehicle collisions are a significant concern in road accidents, particularly with the rise of autonomous driving technology. However, existing studies often struggle to accurately predict collisions due to inconsistent correlations between collected data and collision labels. Therefore, this work quantitatively analyzes traffic accident data and constructs new features with strong correlations to the labels. In this study, a rule classification-dilated convolution network (R-DCN) model, which combines rule learning with dilated convolutional networks, is proposed. The rule learning model predicts partially collided vehicles using predefined rules, resulting in interpretability, high prediction efficiency, and quick computation. The remaining vehicle collisions are estimated using dilated convolutional layers, addressing the issue of missing important features in conventional convolution models. To distinguish between intense collisions (predicted by rule learning) and nonintense collisions (predicted by the dilated convolutional model), the data for training the network are those that remove the intense collision predicted by the rule learning model. The proposed model exhibits enhanced sensitivity to nonintense collision data. Compared to existing models, the approach presented in this work demonstrates superior evaluation metrics and training speed.

Traffic Flow Prediction: A Spatiotemporal Convolution Model Considering Traffic Accident Influence Coefficient Matrix

2023-09-27T21:09:30+00:00

Accurate and timely traffic flow prediction plays a vital role in traffic planning. But the work is more complicated for the road where the traffic accident occurred and the roads around it. This is because the influence of traffic accidents not only acts on the traffic flow of the current road but also quickly spreads to the surrounding road. Traffic accidents differ from other external factors, and their influence on surrounding traffic flow is distinct. Therefore, this work proposes an influence coefficient matrix to express the degree of influence between any two roads to quickly capture the impact of traffic accidents on different road traffic flows. Moreover, this work proposes a hybrid network model based on graph and temporal convolution. To address the spatial dependence between traffic flow data and roads, we selected a graph convolutional network that can be used to analyze the complicated non-Euclidean spatial data in order to extract the spatial dependence. Taking into consideration the temporal dependence of traffic flow data, the temporal convolution model is chosen in this work to model the temporal dependence of the data. Compared to traditional statistical models, single deep learning models, and complex spatiotemporal convolutional models, our model’s performance has been improved by 30% to 50%.

A Deep Spatiotemporal Model for Travel Time Prediction

2023-07-10T22:44:56+00:00

Accurate and reliable travel time prediction is important for promoting the development of urban public transportation, ensuring public travel safety, and establishing smart cities. In travel time prediction, studying spatiotemporal correlation features can help us better understand the dynamic changes and spatial dependencies in traffic data, and it can explain the patterns and trends in vehicle travel. Studying the external factors that influence vehicle travel can help us comprehensively consider the complexity of the transportation system, and incorporating these factors into prediction models can enhance the accuracy and robustness of the models. Therefore, this article proposes a novel deep spatiotemporal model for travel time prediction called DeepSTM-TTP. The architecture of this model consists of three parts: spatial-temporal convolution mechanism, external factor mechanism, and multitask learning mechanism. The spatiotemporal convolution mechanism is used to capture the spatiotemporal correlation of the trajectory external factor mechanism is used to handle the external information in the trajectory. A multitask learning mechanism achieves a balance between local path travel time prediction and whole path travel time prediction. The model fully considers the spatial-temporal correlation of the original GPS location sequence with external information. The experimental results on real datasets demonstrate that the model proposed in this article outperforms four well-known travel time prediction models, including a statistical model (HA), a machine learning model (GBDT), and two deep learning models (DeepTTE and DeepTTE-RNN).

Electrical Equivalent Circuit Modeling of Various Electrically Small Antennas for Biomedical Applications

2021-12-18T08:10:48+00:00

This work outlines the design and development activities of various electrically small antennas for bio-medical applications. It also covers the electrical modeling aspects of all these miniaturized antennas. Three antennas with different specifications have been discussed with diversified proposed applications. First example deals with a single frequency (9.45 GHz) on-chip antenna, whereas the second one covers an ultra-wideband frequency range (2.5 to 20.6 GHz) and finally the third antenna targets an application for 100 GHz band. The size of the first one is 2×2.1 mm², while the second on-chip antenna occupies an area of about 4.6 ×11.5 mm² over silicon substrate. The third antenna module is developed on LCP substrate, which can be accommodated within 12.5×27 mm² area. Though the two on-chip antennas offer only lower gain of around -29 dBi and -3 dBi respectively implementing silicon as a base material, but it paves the way for monolithic integration within a chip. The third candidate exhibits a directive gain of 19-20 dBi with a radiation efficiency of 80% over 100 GHz band. The highlighted portion of this current research work is to propose empirical modeling of electrically small antennas. The proposed methods claim to be most simple in nature and without applying complicated mathematical jugglery easy circuit models are presented for these aforesaid antennas, going to the insight of device physics. A comparative study has been carried out with the proposed model and full-wave simulated results for each antenna, to validate the circuit models.

Two Fluids Cosmological Models in Scale Covariant Theory of Gravitation

2022-03-28T14:23:27+00:00

The present paper deals with Bianchi type I two fluids cosmological model in scale covariant theory of gravitation.Matter fluid modeling observed matter and radiating fluid modeling cosmic microwave background radiation are taken as source. Exact Solutions of the field equations are obtained. Both interacting and non-interacting cases of two fluids are investigated. The exact solutions are obtained for constraints The energy densities are positive for the negative value of parametric constant in case of exponential model. Energy transfer from matter to radiation is observed in case of interacting fluid. Some physical parameter of the obtained model is discussed in detail.

Nonlinear Velocity and Shape Control of a Rotating Smart Flexible Beam-Like Structure

2022-06-14T20:24:56+00:00

Smart materials and structures have a great appeal within the aerospace and automotive communities because they promise to enable better performance and functionality over existing structural and functional materials. The idea proposed in this work is part of this challenging and current scenario and can be applied to the study of morphing wings, helicopter blades, etc. Moreover, the mathematical modeling and velocity and shape control of a rotating flexible beam-like structure are investigated. The nonlinear partial differential governing equations of motion are derived using the extended Hamilton’s Principle and numerically integrated using a combination of finite difference and the fourth-order Runge–Kutta methods. In order to force the flexible structure to assume the desired shape and simultaneously control the velocity of the rotating axis, the optimal nonlinear control method named state-dependent Riccati equation (SDRE) is considered. This control technique is applied to piezoelectric actuators along the beam and an external torque coming from a DC motor and acting in the rotating axis. The numerical simulation results show that the proposed control technique is efficient when acting along the rotating beam to deform it into the desired shape while also acting on the motor axis to keep the rotation speed constant.

On an Inverse Problem of Identifying an Unknown Boundary for the Biharmonic Equation from Cauchy Data

2022-01-20T16:30:50+00:00

In this paper, we deal with an inverse problem for the biharmonic equation to find an unknown boundary in the plane by using an additional information assumed on the remaining known part of the boundary. As a by-product, we can uniquely determine the solution everywhere in its domain of definition by supposing that the available data have Fourier expansions. The question of the existence and uniqueness of this inverse problem will be investigated, and we will conclude with some analytical examples to ensure the validity of this study.

Quattuortrigintic Functional Equation and its Hyers-Ulam Stability

2022-12-03T19:30:50+00:00

This article deals a new Quattuortrigintic functional equation with the general solution and the generalized Hyers-Ulam stability in multi-Banach space and menger probabilistic normed space by employing fixed point technique.